test_callgraph_cache.py
python
sha256:51ce277f663e01a43eaffbe77509b1de7ac2d4251b55d23306304bcdeb92c90d
feat(pack): delta-encode snapshots in MPackBundle wire format
Sonnet 4.6
minor
⚠ breaking
121 days ago
| 1 | """TDD tests for CallGraphCache — persistent per-file call-graph cache. |
| 2 | |
| 3 | Architecture |
| 4 | ------------ |
| 5 | ``build_forward_graph`` re-reads every Python blob from the object store and |
| 6 | re-parses every AST on every CLI invocation. For a 778-file Python codebase |
| 7 | that costs ~15 s cold; with a warm cache it should be ~1 s. |
| 8 | |
| 9 | ``CallGraphCache`` mirrors ``SymbolCache`` exactly, but stores a per-file |
| 10 | **subgraph** rather than a symbol tree. The subgraph is the portion of the |
| 11 | forward call graph contributed by one source file: |
| 12 | |
| 13 | ``{caller_address: frozenset[callee_bare_name]}`` |
| 14 | |
| 15 | This preserves full per-address granularity so the warm path produces an |
| 16 | identical ``ForwardGraph`` to the cold path. |
| 17 | |
| 18 | Key: SHA-256 of file bytes (``object_id`` from the manifest). |
| 19 | Value: ``dict[str, frozenset[str]]`` — per-file forward subgraph. |
| 20 | File: ``.muse/cache/callgraph.msgpack`` (msgpack, atomic write). |
| 21 | API: ``load(muse_dir)`` / ``empty()`` / ``get`` / ``put`` / ``prune`` / |
| 22 | ``size`` / ``save`` + convenience ``load_callgraph_cache(root)``. |
| 23 | |
| 24 | Coverage matrix |
| 25 | --------------- |
| 26 | - Memory operations: get/put/dirty/size/prune/empty class method |
| 27 | - Persistence: save creates file; save/load round-trip; no-dirty skip; |
| 28 | dirty=False after save; second save is no-op; atomic write |
| 29 | - Graceful load: missing file → empty; corrupt → empty; wrong version → empty; |
| 30 | invalid entry skipped, valid entries survive |
| 31 | - Convenience helper: load_callgraph_cache with and without .muse dir |
| 32 | - Integration cold: build_forward_graph with callgraph_cache=None produces correct graph |
| 33 | - Integration warm: pre-populated cache skips read_object and ast.parse entirely |
| 34 | - Integration save: cache entries populated after build; build does NOT auto-save |
| 35 | - Non-Python files: not cached, not added to graph |
| 36 | - Syntax errors: parse failure → entry not cached |
| 37 | - Correctness: warm graph == cold graph for all addresses |
| 38 | - Performance: warm call ≥ 5× faster than cold; < 100 ms for 30 files |
| 39 | """ |
| 40 | |
| 41 | from __future__ import annotations |
| 42 | |
| 43 | import pathlib |
| 44 | import time |
| 45 | from unittest.mock import patch |
| 46 | |
| 47 | import msgpack |
| 48 | import pytest |
| 49 | |
| 50 | from muse.core.object_store import write_object |
| 51 | from muse.core.paths import muse_dir |
| 52 | from muse.core.types import Manifest, blob_id |
| 53 | |
| 54 | # Subgraph type alias: the per-file portion of ForwardGraph |
| 55 | _Subgraph = dict[str, frozenset[str]] |
| 56 | |
| 57 | |
| 58 | # --------------------------------------------------------------------------- |
| 59 | # Helpers — shared across all test classes |
| 60 | # --------------------------------------------------------------------------- |
| 61 | |
| 62 | |
| 63 | def _muse_dir(tmp_path: pathlib.Path) -> pathlib.Path: |
| 64 | d = muse_dir(tmp_path) |
| 65 | d.mkdir(exist_ok=True) |
| 66 | (d / "cache").mkdir(exist_ok=True) |
| 67 | return d |
| 68 | |
| 69 | |
| 70 | def _write_py( |
| 71 | tmp_path: pathlib.Path, rel_path: str, source: str |
| 72 | ) -> tuple[str, bytes]: |
| 73 | """Write source to object store; return (object_id, raw_bytes). |
| 74 | |
| 75 | object_id uses the canonical ``sha256:<hex>`` prefix required by the |
| 76 | object store's ``validate_object_id`` guard. |
| 77 | """ |
| 78 | raw = source.encode() |
| 79 | oid = blob_id(raw) |
| 80 | write_object(tmp_path, oid, raw) |
| 81 | return oid, raw |
| 82 | |
| 83 | |
| 84 | def _make_manifest( |
| 85 | tmp_path: pathlib.Path, files: dict[str, str] |
| 86 | ) -> Manifest: |
| 87 | manifest: Manifest = {} |
| 88 | for rel_path, source in files.items(): |
| 89 | oid, _ = _write_py(tmp_path, rel_path, source) |
| 90 | manifest[rel_path] = oid |
| 91 | return manifest |
| 92 | |
| 93 | |
| 94 | def _subgraph( |
| 95 | file_path: str, caller: str, callees: set[str] |
| 96 | ) -> _Subgraph: |
| 97 | """Helper to build a minimal per-file subgraph for a single caller.""" |
| 98 | return {f"{file_path}::{caller}": frozenset(callees)} |
| 99 | |
| 100 | |
| 101 | # --------------------------------------------------------------------------- |
| 102 | # TestCallGraphCacheMemory |
| 103 | # --------------------------------------------------------------------------- |
| 104 | |
| 105 | |
| 106 | class TestCallGraphCacheMemory: |
| 107 | """In-memory get/put/prune/size/empty operations.""" |
| 108 | |
| 109 | def test_empty_get_miss(self) -> None: |
| 110 | from muse.core.callgraph_cache import CallGraphCache |
| 111 | cache = CallGraphCache.empty() |
| 112 | assert cache.get("no_such_id") is None |
| 113 | |
| 114 | def test_put_then_get_hit(self) -> None: |
| 115 | from muse.core.callgraph_cache import CallGraphCache |
| 116 | cache = CallGraphCache.empty() |
| 117 | subgraph: _Subgraph = { |
| 118 | "mod.py::caller": frozenset({"compute", "validate"}), |
| 119 | "mod.py::leaf": frozenset(), |
| 120 | } |
| 121 | cache.put("abc123", subgraph) |
| 122 | assert cache.get("abc123") == subgraph |
| 123 | |
| 124 | def test_put_marks_dirty(self) -> None: |
| 125 | from muse.core.callgraph_cache import CallGraphCache |
| 126 | cache = CallGraphCache.empty() |
| 127 | assert not cache._dirty |
| 128 | cache.put("id1", {"mod.py::fn": frozenset()}) |
| 129 | assert cache._dirty |
| 130 | |
| 131 | def test_different_ids_independent(self) -> None: |
| 132 | from muse.core.callgraph_cache import CallGraphCache |
| 133 | cache = CallGraphCache.empty() |
| 134 | sg_a: _Subgraph = {"a.py::alpha": frozenset({"beta"})} |
| 135 | sg_b: _Subgraph = {"b.py::gamma": frozenset({"delta"})} |
| 136 | cache.put("id_a", sg_a) |
| 137 | cache.put("id_b", sg_b) |
| 138 | assert cache.get("id_a") == sg_a |
| 139 | assert cache.get("id_b") == sg_b |
| 140 | |
| 141 | def test_put_same_id_overwrites(self) -> None: |
| 142 | from muse.core.callgraph_cache import CallGraphCache |
| 143 | cache = CallGraphCache.empty() |
| 144 | cache.put("same", {"f.py::old": frozenset({"x"})}) |
| 145 | cache.put("same", {"f.py::new": frozenset({"y"})}) |
| 146 | result = cache.get("same") |
| 147 | assert "f.py::new" in result |
| 148 | assert "f.py::old" not in result |
| 149 | assert cache.size == 1 |
| 150 | |
| 151 | def test_size_starts_zero(self) -> None: |
| 152 | from muse.core.callgraph_cache import CallGraphCache |
| 153 | assert CallGraphCache.empty().size == 0 |
| 154 | |
| 155 | def test_size_grows_with_put(self) -> None: |
| 156 | from muse.core.callgraph_cache import CallGraphCache |
| 157 | cache = CallGraphCache.empty() |
| 158 | cache.put("x", {"m.py::f": frozenset()}) |
| 159 | cache.put("y", {"m.py::g": frozenset()}) |
| 160 | assert cache.size == 2 |
| 161 | |
| 162 | def test_prune_removes_stale_entries(self) -> None: |
| 163 | from muse.core.callgraph_cache import CallGraphCache |
| 164 | cache = CallGraphCache.empty() |
| 165 | cache.put("keep", {"a.py::f": frozenset()}) |
| 166 | cache.put("drop", {"b.py::g": frozenset()}) |
| 167 | cache.prune({"keep"}) |
| 168 | assert cache.get("keep") is not None |
| 169 | assert cache.get("drop") is None |
| 170 | |
| 171 | def test_prune_marks_dirty_when_stale(self) -> None: |
| 172 | from muse.core.callgraph_cache import CallGraphCache |
| 173 | cache = CallGraphCache.empty() |
| 174 | cache.put("drop", {"a.py::f": frozenset()}) |
| 175 | cache._dirty = False |
| 176 | cache.prune(set()) |
| 177 | assert cache._dirty |
| 178 | |
| 179 | def test_prune_no_stale_not_dirty(self) -> None: |
| 180 | from muse.core.callgraph_cache import CallGraphCache |
| 181 | cache = CallGraphCache.empty() |
| 182 | cache.put("keep", {"a.py::f": frozenset()}) |
| 183 | cache._dirty = False |
| 184 | cache.prune({"keep", "other"}) |
| 185 | assert not cache._dirty |
| 186 | |
| 187 | def test_empty_save_is_noop(self, tmp_path: pathlib.Path) -> None: |
| 188 | from muse.core.callgraph_cache import CallGraphCache |
| 189 | cache = CallGraphCache.empty() |
| 190 | cache.put("id", {"m.py::fn": frozenset({"fn"})}) |
| 191 | cache.save() # muse_dir is None — must not raise |
| 192 | assert not any(tmp_path.rglob("callgraph.msgpack")) |
| 193 | |
| 194 | def test_empty_subgraph_is_valid(self) -> None: |
| 195 | from muse.core.callgraph_cache import CallGraphCache |
| 196 | cache = CallGraphCache.empty() |
| 197 | cache.put("leaf_fn", {}) |
| 198 | result = cache.get("leaf_fn") |
| 199 | assert result == {} |
| 200 | |
| 201 | def test_frozensets_preserved_in_subgraph(self) -> None: |
| 202 | from muse.core.callgraph_cache import CallGraphCache |
| 203 | cache = CallGraphCache.empty() |
| 204 | sg: _Subgraph = {"m.py::fn": frozenset({"a", "b", "c"})} |
| 205 | cache.put("id", sg) |
| 206 | result = cache.get("id") |
| 207 | assert isinstance(result["m.py::fn"], frozenset) |
| 208 | assert result["m.py::fn"] == frozenset({"a", "b", "c"}) |
| 209 | |
| 210 | |
| 211 | # --------------------------------------------------------------------------- |
| 212 | # TestCallGraphCachePersistence |
| 213 | # --------------------------------------------------------------------------- |
| 214 | |
| 215 | |
| 216 | class TestCallGraphCachePersistence: |
| 217 | """save() / load() round-trip via .muse/cache/callgraph.msgpack.""" |
| 218 | |
| 219 | def test_save_creates_file(self, tmp_path: pathlib.Path) -> None: |
| 220 | from muse.core.callgraph_cache import CallGraphCache |
| 221 | md = _muse_dir(tmp_path) |
| 222 | cache = CallGraphCache.load(md) |
| 223 | cache.put("id1", {"m.py::fn": frozenset({"compute"})}) |
| 224 | cache.save() |
| 225 | assert (md / "cache" / "callgraph.msgpack").is_file() |
| 226 | |
| 227 | def test_save_then_load_round_trip(self, tmp_path: pathlib.Path) -> None: |
| 228 | from muse.core.callgraph_cache import CallGraphCache |
| 229 | md = _muse_dir(tmp_path) |
| 230 | sg: _Subgraph = { |
| 231 | "billing.py::compute": frozenset({"validate", "send"}), |
| 232 | "billing.py::validate": frozenset(), |
| 233 | } |
| 234 | oid = "deadbeef" * 8 |
| 235 | |
| 236 | cache = CallGraphCache.load(md) |
| 237 | cache.put(oid, sg) |
| 238 | cache.save() |
| 239 | |
| 240 | loaded = CallGraphCache.load(md) |
| 241 | result = loaded.get(oid) |
| 242 | assert result is not None |
| 243 | assert result == sg |
| 244 | |
| 245 | def test_round_trip_preserves_frozenset_type(self, tmp_path: pathlib.Path) -> None: |
| 246 | from muse.core.callgraph_cache import CallGraphCache |
| 247 | md = _muse_dir(tmp_path) |
| 248 | cache = CallGraphCache.load(md) |
| 249 | cache.put("id", {"m.py::fn": frozenset({"a", "b"})}) |
| 250 | cache.save() |
| 251 | |
| 252 | loaded = CallGraphCache.load(md) |
| 253 | result = loaded.get("id") |
| 254 | assert isinstance(result["m.py::fn"], frozenset) |
| 255 | |
| 256 | def test_save_no_dirty_skips_write(self, tmp_path: pathlib.Path) -> None: |
| 257 | from muse.core.callgraph_cache import CallGraphCache |
| 258 | md = _muse_dir(tmp_path) |
| 259 | cache = CallGraphCache.load(md) |
| 260 | cache.save() |
| 261 | assert not (md / "cache" / "callgraph.msgpack").is_file() |
| 262 | |
| 263 | def test_save_clears_dirty_flag(self, tmp_path: pathlib.Path) -> None: |
| 264 | from muse.core.callgraph_cache import CallGraphCache |
| 265 | md = _muse_dir(tmp_path) |
| 266 | cache = CallGraphCache.load(md) |
| 267 | cache.put("id", {"m.py::fn": frozenset()}) |
| 268 | cache.save() |
| 269 | assert not cache._dirty |
| 270 | |
| 271 | def test_second_save_is_noop(self, tmp_path: pathlib.Path) -> None: |
| 272 | from muse.core.callgraph_cache import CallGraphCache |
| 273 | md = _muse_dir(tmp_path) |
| 274 | cache = CallGraphCache.load(md) |
| 275 | cache.put("id", {"m.py::fn": frozenset()}) |
| 276 | cache.save() |
| 277 | mtime1 = (md / "cache" / "callgraph.msgpack").stat().st_mtime_ns |
| 278 | cache.save() |
| 279 | mtime2 = (md / "cache" / "callgraph.msgpack").stat().st_mtime_ns |
| 280 | assert mtime1 == mtime2 |
| 281 | |
| 282 | def test_multiple_entries_survive_round_trip(self, tmp_path: pathlib.Path) -> None: |
| 283 | from muse.core.callgraph_cache import CallGraphCache |
| 284 | md = _muse_dir(tmp_path) |
| 285 | cache = CallGraphCache.load(md) |
| 286 | entries: dict[str, _Subgraph] = { |
| 287 | "id_a": {"a.py::alpha": frozenset({"beta"})}, |
| 288 | "id_b": {}, |
| 289 | "id_c": {"c.py::gamma": frozenset({"delta", "epsilon"})}, |
| 290 | } |
| 291 | for oid, sg in entries.items(): |
| 292 | cache.put(oid, sg) |
| 293 | cache.save() |
| 294 | |
| 295 | loaded = CallGraphCache.load(md) |
| 296 | for oid, sg in entries.items(): |
| 297 | assert loaded.get(oid) == sg |
| 298 | |
| 299 | def test_atomic_write_no_tmp_leftover(self, tmp_path: pathlib.Path) -> None: |
| 300 | from muse.core.callgraph_cache import CallGraphCache |
| 301 | md = _muse_dir(tmp_path) |
| 302 | cache = CallGraphCache.load(md) |
| 303 | cache.put("id", {"m.py::fn": frozenset()}) |
| 304 | cache.save() |
| 305 | assert not any((md / "cache").glob("*.tmp")) |
| 306 | |
| 307 | def test_orphaned_tmp_swept_on_startup(self, tmp_path: pathlib.Path) -> None: |
| 308 | """A stale ``.callgraph_*.tmp`` left by a crash is removed by the startup sweep.""" |
| 309 | from muse.core.repo import _cleanup_muse_dir_temps |
| 310 | md = _muse_dir(tmp_path) |
| 311 | orphan = md / "cache" / ".callgraph_abc123.tmp" |
| 312 | orphan.write_bytes(b"stale") |
| 313 | _cleanup_muse_dir_temps(md) |
| 314 | assert not orphan.exists() |
| 315 | |
| 316 | |
| 317 | # --------------------------------------------------------------------------- |
| 318 | # TestCallGraphCacheGracefulLoad |
| 319 | # --------------------------------------------------------------------------- |
| 320 | |
| 321 | |
| 322 | class TestCallGraphCacheGracefulLoad: |
| 323 | """load() never raises — returns empty cache on any error.""" |
| 324 | |
| 325 | def test_absent_file_returns_empty(self, tmp_path: pathlib.Path) -> None: |
| 326 | from muse.core.callgraph_cache import CallGraphCache |
| 327 | md = _muse_dir(tmp_path) |
| 328 | assert CallGraphCache.load(md).size == 0 |
| 329 | |
| 330 | def test_corrupt_file_returns_empty(self, tmp_path: pathlib.Path) -> None: |
| 331 | from muse.core.callgraph_cache import CallGraphCache |
| 332 | md = _muse_dir(tmp_path) |
| 333 | (md / "cache" / "callgraph.msgpack").write_bytes(b"not valid msgpack !!!") |
| 334 | assert CallGraphCache.load(md).size == 0 |
| 335 | |
| 336 | def test_wrong_version_returns_empty(self, tmp_path: pathlib.Path) -> None: |
| 337 | from muse.core.callgraph_cache import CallGraphCache |
| 338 | md = _muse_dir(tmp_path) |
| 339 | doc = {"version": 999, "entries": {}} |
| 340 | (md / "cache" / "callgraph.msgpack").write_bytes(msgpack.packb(doc, use_bin_type=True)) |
| 341 | assert CallGraphCache.load(md).size == 0 |
| 342 | |
| 343 | def test_non_dict_entries_returns_empty(self, tmp_path: pathlib.Path) -> None: |
| 344 | from muse.core.callgraph_cache import CallGraphCache |
| 345 | md = _muse_dir(tmp_path) |
| 346 | doc = {"version": 1, "entries": "not a dict"} |
| 347 | (md / "cache" / "callgraph.msgpack").write_bytes(msgpack.packb(doc, use_bin_type=True)) |
| 348 | assert CallGraphCache.load(md).size == 0 |
| 349 | |
| 350 | def test_invalid_entry_skipped_valid_survive(self, tmp_path: pathlib.Path) -> None: |
| 351 | """A single malformed subgraph entry is skipped; valid ones survive.""" |
| 352 | from muse.core.callgraph_cache import CallGraphCache |
| 353 | md = _muse_dir(tmp_path) |
| 354 | doc = { |
| 355 | "version": 1, |
| 356 | "entries": { |
| 357 | "good_id": {"m.py::fn": ["compute", "validate"]}, # valid |
| 358 | "bad_id": {"m.py::fn": "not_a_list"}, # invalid — str not list |
| 359 | "empty_id": {}, # valid empty subgraph |
| 360 | }, |
| 361 | } |
| 362 | (md / "cache" / "callgraph.msgpack").write_bytes(msgpack.packb(doc, use_bin_type=True)) |
| 363 | cache = CallGraphCache.load(md) |
| 364 | assert cache.get("good_id") == {"m.py::fn": frozenset({"compute", "validate"})} |
| 365 | assert cache.get("bad_id") is None |
| 366 | assert cache.get("empty_id") == {} |
| 367 | |
| 368 | def test_entry_with_non_str_callee_skipped(self, tmp_path: pathlib.Path) -> None: |
| 369 | from muse.core.callgraph_cache import CallGraphCache |
| 370 | md = _muse_dir(tmp_path) |
| 371 | doc = { |
| 372 | "version": 1, |
| 373 | "entries": { |
| 374 | "bad": {"m.py::fn": [123, "valid"]}, # 123 is not str → skip entry |
| 375 | }, |
| 376 | } |
| 377 | (md / "cache" / "callgraph.msgpack").write_bytes(msgpack.packb(doc, use_bin_type=True)) |
| 378 | assert CallGraphCache.load(md).get("bad") is None |
| 379 | |
| 380 | def test_entry_subgraph_not_dict_skipped(self, tmp_path: pathlib.Path) -> None: |
| 381 | from muse.core.callgraph_cache import CallGraphCache |
| 382 | md = _muse_dir(tmp_path) |
| 383 | doc = { |
| 384 | "version": 1, |
| 385 | "entries": { |
| 386 | "bad": "not_a_dict_at_all", # subgraph must be a dict |
| 387 | }, |
| 388 | } |
| 389 | (md / "cache" / "callgraph.msgpack").write_bytes(msgpack.packb(doc, use_bin_type=True)) |
| 390 | assert CallGraphCache.load(md).get("bad") is None |
| 391 | |
| 392 | |
| 393 | # --------------------------------------------------------------------------- |
| 394 | # TestLoadCallGraphCache — convenience helper |
| 395 | # --------------------------------------------------------------------------- |
| 396 | |
| 397 | |
| 398 | class TestLoadCallGraphCache: |
| 399 | """load_callgraph_cache(root) convenience loader.""" |
| 400 | |
| 401 | def test_no_muse_dir_returns_empty(self, tmp_path: pathlib.Path) -> None: |
| 402 | from muse.core.callgraph_cache import load_callgraph_cache |
| 403 | assert load_callgraph_cache(tmp_path).size == 0 |
| 404 | |
| 405 | def test_with_muse_dir_loads_existing(self, tmp_path: pathlib.Path) -> None: |
| 406 | from muse.core.callgraph_cache import CallGraphCache, load_callgraph_cache |
| 407 | md = _muse_dir(tmp_path) |
| 408 | seed = CallGraphCache.load(md) |
| 409 | seed.put("myid", {"m.py::fn": frozenset({"fn"})}) |
| 410 | seed.save() |
| 411 | |
| 412 | cache = load_callgraph_cache(tmp_path) |
| 413 | assert cache.get("myid") == {"m.py::fn": frozenset({"fn"})} |
| 414 | |
| 415 | def test_with_empty_muse_dir_returns_empty(self, tmp_path: pathlib.Path) -> None: |
| 416 | from muse.core.callgraph_cache import load_callgraph_cache |
| 417 | _muse_dir(tmp_path) |
| 418 | assert load_callgraph_cache(tmp_path).size == 0 |
| 419 | |
| 420 | |
| 421 | # --------------------------------------------------------------------------- |
| 422 | # TestBuildForwardGraphWithCache — integration |
| 423 | # --------------------------------------------------------------------------- |
| 424 | |
| 425 | |
| 426 | class TestBuildForwardGraphWithCache: |
| 427 | """build_forward_graph accepts an optional CallGraphCache and uses it.""" |
| 428 | |
| 429 | def _repo(self, tmp_path: pathlib.Path) -> pathlib.Path: |
| 430 | _muse_dir(tmp_path) |
| 431 | return tmp_path |
| 432 | |
| 433 | def test_cold_cache_none_correct_graph(self, tmp_path: pathlib.Path) -> None: |
| 434 | """Without a cache, build_forward_graph still returns the correct graph.""" |
| 435 | root = self._repo(tmp_path) |
| 436 | src = "def caller():\n callee()\n\ndef callee():\n pass\n" |
| 437 | manifest = _make_manifest(root, {"mod.py": src}) |
| 438 | |
| 439 | from muse.plugins.code._callgraph import build_forward_graph |
| 440 | graph = build_forward_graph(root, manifest) |
| 441 | caller_addr = next(k for k in graph if "caller" in k) |
| 442 | assert "callee" in graph[caller_addr] |
| 443 | |
| 444 | def test_explicit_cache_hit_skips_read_object(self, tmp_path: pathlib.Path) -> None: |
| 445 | """When the cache has an entry for object_id, read_object is not called.""" |
| 446 | from muse.core.callgraph_cache import CallGraphCache |
| 447 | from muse.plugins.code._callgraph import build_forward_graph |
| 448 | |
| 449 | root = self._repo(tmp_path) |
| 450 | src = "def caller():\n callee()\n" |
| 451 | oid, _ = _write_py(root, "mod.py", src) |
| 452 | manifest = {"mod.py": oid} |
| 453 | |
| 454 | # Pre-populate with the known subgraph |
| 455 | cache = CallGraphCache.empty() |
| 456 | cache.put(oid, {"mod.py::caller": frozenset({"callee"})}) |
| 457 | |
| 458 | with patch("muse.plugins.code._callgraph.read_object") as mock_read: |
| 459 | build_forward_graph(root, manifest, callgraph_cache=cache) |
| 460 | mock_read.assert_not_called() |
| 461 | |
| 462 | def test_cache_miss_calls_read_object(self, tmp_path: pathlib.Path) -> None: |
| 463 | """On a cache miss, read_object IS called (the normal cold path).""" |
| 464 | from muse.core.callgraph_cache import CallGraphCache |
| 465 | from muse.plugins.code._callgraph import build_forward_graph |
| 466 | |
| 467 | root = self._repo(tmp_path) |
| 468 | src = "def caller():\n callee()\n" |
| 469 | oid, _ = _write_py(root, "mod.py", src) |
| 470 | manifest = {"mod.py": oid} |
| 471 | |
| 472 | empty_cache = CallGraphCache.empty() |
| 473 | with patch( |
| 474 | "muse.plugins.code._callgraph.read_object", |
| 475 | wraps=__import__( |
| 476 | "muse.core.object_store", fromlist=["read_object"] |
| 477 | ).read_object, |
| 478 | ) as mock_read: |
| 479 | build_forward_graph(root, manifest, callgraph_cache=empty_cache) |
| 480 | assert mock_read.call_count >= 1 |
| 481 | |
| 482 | def test_cache_populated_after_build(self, tmp_path: pathlib.Path) -> None: |
| 483 | """After build_forward_graph, the cache contains a subgraph for each parsed file.""" |
| 484 | from muse.core.callgraph_cache import CallGraphCache |
| 485 | from muse.plugins.code._callgraph import build_forward_graph |
| 486 | |
| 487 | root = self._repo(tmp_path) |
| 488 | src = "def caller():\n callee()\n" |
| 489 | oid, _ = _write_py(root, "mod.py", src) |
| 490 | manifest = {"mod.py": oid} |
| 491 | |
| 492 | cache = CallGraphCache.empty() |
| 493 | build_forward_graph(root, manifest, callgraph_cache=cache) |
| 494 | result = cache.get(oid) |
| 495 | assert result is not None |
| 496 | assert isinstance(result, dict) |
| 497 | |
| 498 | def test_cache_subgraph_has_correct_callees(self, tmp_path: pathlib.Path) -> None: |
| 499 | """The subgraph stored in the cache matches the cold graph output.""" |
| 500 | from muse.core.callgraph_cache import CallGraphCache |
| 501 | from muse.plugins.code._callgraph import build_forward_graph |
| 502 | |
| 503 | root = self._repo(tmp_path) |
| 504 | src = "def caller():\n callee_a()\n callee_b()\n" |
| 505 | oid, _ = _write_py(root, "mod.py", src) |
| 506 | manifest = {"mod.py": oid} |
| 507 | |
| 508 | cache = CallGraphCache.empty() |
| 509 | graph = build_forward_graph(root, manifest, callgraph_cache=cache) |
| 510 | |
| 511 | # The subgraph in the cache must match the graph output for this file |
| 512 | subgraph = cache.get(oid) |
| 513 | assert subgraph is not None |
| 514 | for addr, callees in graph.items(): |
| 515 | assert addr in subgraph |
| 516 | assert subgraph[addr] == callees |
| 517 | |
| 518 | def test_warm_graph_equals_cold_graph(self, tmp_path: pathlib.Path) -> None: |
| 519 | """Graph built with a warm cache equals graph built cold.""" |
| 520 | from muse.core.callgraph_cache import CallGraphCache |
| 521 | from muse.plugins.code._callgraph import build_forward_graph |
| 522 | |
| 523 | root = self._repo(tmp_path) |
| 524 | src = "def a():\n b()\n c()\n\ndef b():\n pass\n\ndef c():\n pass\n" |
| 525 | oid, _ = _write_py(root, "mod.py", src) |
| 526 | manifest = {"mod.py": oid} |
| 527 | |
| 528 | cold_graph = build_forward_graph(root, manifest) |
| 529 | |
| 530 | cache = CallGraphCache.empty() |
| 531 | build_forward_graph(root, manifest, callgraph_cache=cache) # populates cache |
| 532 | warm_graph = build_forward_graph(root, manifest, callgraph_cache=cache) # warm |
| 533 | |
| 534 | for addr in cold_graph: |
| 535 | assert addr in warm_graph |
| 536 | assert cold_graph[addr] == warm_graph[addr] |
| 537 | |
| 538 | def test_non_python_files_not_cached(self, tmp_path: pathlib.Path) -> None: |
| 539 | """Non-Python files are not added to the cache or graph.""" |
| 540 | from muse.core.callgraph_cache import CallGraphCache |
| 541 | from muse.plugins.code._callgraph import build_forward_graph |
| 542 | |
| 543 | root = self._repo(tmp_path) |
| 544 | md_oid, _ = _write_py(root, "README.md", "# My README\n") |
| 545 | py_oid, _ = _write_py(root, "mod.py", "def fn():\n pass\n") |
| 546 | manifest = {"README.md": md_oid, "mod.py": py_oid} |
| 547 | |
| 548 | cache = CallGraphCache.empty() |
| 549 | build_forward_graph(root, manifest, callgraph_cache=cache) |
| 550 | |
| 551 | assert cache.get(md_oid) is None # markdown — not cached |
| 552 | assert cache.get(py_oid) is not None # Python — cached |
| 553 | |
| 554 | def test_build_does_not_call_cache_save(self, tmp_path: pathlib.Path) -> None: |
| 555 | """build_forward_graph must not call cache.save() — that is the caller's job.""" |
| 556 | from muse.core.callgraph_cache import CallGraphCache |
| 557 | from muse.plugins.code._callgraph import build_forward_graph |
| 558 | |
| 559 | root = self._repo(tmp_path) |
| 560 | src = "def fn():\n pass\n" |
| 561 | oid, _ = _write_py(root, "mod.py", src) |
| 562 | manifest = {"mod.py": oid} |
| 563 | |
| 564 | md = _muse_dir(tmp_path) |
| 565 | cache = CallGraphCache.load(md) |
| 566 | build_forward_graph(root, manifest, callgraph_cache=cache) |
| 567 | # If build_forward_graph called save(), the file would exist now |
| 568 | assert not (md / "cache" / "callgraph.msgpack").is_file() |
| 569 | |
| 570 | def test_syntax_error_file_not_cached(self, tmp_path: pathlib.Path) -> None: |
| 571 | """Files with syntax errors are skipped — nothing added to cache.""" |
| 572 | from muse.core.callgraph_cache import CallGraphCache |
| 573 | from muse.plugins.code._callgraph import build_forward_graph |
| 574 | |
| 575 | root = self._repo(tmp_path) |
| 576 | bad_src = "def broken(:\n pass\n" |
| 577 | oid, _ = _write_py(root, "broken.py", bad_src) |
| 578 | manifest = {"broken.py": oid} |
| 579 | |
| 580 | cache = CallGraphCache.empty() |
| 581 | build_forward_graph(root, manifest, callgraph_cache=cache) |
| 582 | assert cache.get(oid) is None # parse failed → not cached |
| 583 | |
| 584 | def test_second_call_skips_ast_parse(self, tmp_path: pathlib.Path) -> None: |
| 585 | """On the second call with a warm cache, ast.parse is never called.""" |
| 586 | from muse.core.callgraph_cache import CallGraphCache |
| 587 | from muse.plugins.code._callgraph import build_forward_graph |
| 588 | |
| 589 | root = self._repo(tmp_path) |
| 590 | src = "def caller():\n callee()\n" |
| 591 | oid, _ = _write_py(root, "mod.py", src) |
| 592 | manifest = {"mod.py": oid} |
| 593 | |
| 594 | cache = CallGraphCache.empty() |
| 595 | build_forward_graph(root, manifest, callgraph_cache=cache) # cold |
| 596 | |
| 597 | with patch("muse.plugins.code._callgraph.ast") as mock_ast: |
| 598 | build_forward_graph(root, manifest, callgraph_cache=cache) # warm |
| 599 | mock_ast.parse.assert_not_called() |
| 600 | |
| 601 | |
| 602 | # --------------------------------------------------------------------------- |
| 603 | # TestCallGraphCachePerformance |
| 604 | # --------------------------------------------------------------------------- |
| 605 | |
| 606 | |
| 607 | class TestCallGraphCachePerformance: |
| 608 | """Second call with a warm cache must be substantially faster than cold.""" |
| 609 | |
| 610 | def _build_repo( |
| 611 | self, tmp_path: pathlib.Path, n_files: int = 30 |
| 612 | ) -> tuple[pathlib.Path, Manifest]: |
| 613 | root = tmp_path |
| 614 | _muse_dir(root) |
| 615 | manifest: Manifest = {} |
| 616 | for i in range(n_files): |
| 617 | src = ( |
| 618 | f"def fn_{i}():\n helper_{i}()\n util_{i}()\n\n" |
| 619 | f"def helper_{i}():\n pass\n\n" |
| 620 | f"def util_{i}():\n pass\n" |
| 621 | ) |
| 622 | oid, _ = _write_py(root, f"mod_{i}.py", src) |
| 623 | manifest[f"mod_{i}.py"] = oid |
| 624 | return root, manifest |
| 625 | |
| 626 | def test_warm_cache_at_least_5x_faster(self, tmp_path: pathlib.Path) -> None: |
| 627 | from muse.core.callgraph_cache import CallGraphCache |
| 628 | from muse.plugins.code._callgraph import build_forward_graph |
| 629 | |
| 630 | root, manifest = self._build_repo(tmp_path, n_files=30) |
| 631 | |
| 632 | cache = CallGraphCache.empty() |
| 633 | t0 = time.perf_counter() |
| 634 | build_forward_graph(root, manifest, callgraph_cache=cache) |
| 635 | cold_ms = (time.perf_counter() - t0) * 1000 |
| 636 | |
| 637 | t1 = time.perf_counter() |
| 638 | build_forward_graph(root, manifest, callgraph_cache=cache) |
| 639 | warm_ms = (time.perf_counter() - t1) * 1000 |
| 640 | |
| 641 | assert warm_ms < cold_ms / 5, ( |
| 642 | f"Warm ({warm_ms:.1f} ms) should be ≥5× faster than cold ({cold_ms:.1f} ms)" |
| 643 | ) |
| 644 | |
| 645 | def test_warm_cache_under_100ms_for_30_files(self, tmp_path: pathlib.Path) -> None: |
| 646 | from muse.core.callgraph_cache import CallGraphCache |
| 647 | from muse.plugins.code._callgraph import build_forward_graph |
| 648 | |
| 649 | root, manifest = self._build_repo(tmp_path, n_files=30) |
| 650 | |
| 651 | cache = CallGraphCache.empty() |
| 652 | build_forward_graph(root, manifest, callgraph_cache=cache) # warm the cache |
| 653 | |
| 654 | t0 = time.perf_counter() |
| 655 | build_forward_graph(root, manifest, callgraph_cache=cache) |
| 656 | warm_ms = (time.perf_counter() - t0) * 1000 |
| 657 | |
| 658 | assert warm_ms < 100, f"Warm call took {warm_ms:.1f} ms — expected < 100 ms" |
| 659 | |
| 660 | def test_graph_correctness_not_degraded_by_cache(self, tmp_path: pathlib.Path) -> None: |
| 661 | """Warm-cache graph is identical to cold-cache graph for all addresses.""" |
| 662 | from muse.core.callgraph_cache import CallGraphCache |
| 663 | from muse.plugins.code._callgraph import build_forward_graph |
| 664 | |
| 665 | root, manifest = self._build_repo(tmp_path, n_files=10) |
| 666 | |
| 667 | cold_graph = build_forward_graph(root, manifest) |
| 668 | |
| 669 | cache = CallGraphCache.empty() |
| 670 | build_forward_graph(root, manifest, callgraph_cache=cache) # warm |
| 671 | warm_graph = build_forward_graph(root, manifest, callgraph_cache=cache) |
| 672 | |
| 673 | for addr in cold_graph: |
| 674 | assert addr in warm_graph, f"Address {addr!r} missing from warm graph" |
| 675 | assert cold_graph[addr] == warm_graph[addr], ( |
| 676 | f"Callee mismatch at {addr!r}: " |
| 677 | f"cold={cold_graph[addr]} warm={warm_graph[addr]}" |
| 678 | ) |
File History
1 commit
sha256:51ce277f663e01a43eaffbe77509b1de7ac2d4251b55d23306304bcdeb92c90d
feat(pack): delta-encode snapshots in MPackBundle wire format
Sonnet 4.6
minor
⚠
121 days ago